Predicting the stock market with big-data analysis of tweets
Articles Blog

Predicting the stock market with big-data analysis of tweets


Can tweets about the stock market be
ignored as random noise, generated by self-proclaimed trading gurus? A study by a team of researchers at Rotterdam School of Management, Erasmus University now shows that tweets contain useful information that can be used to predict
stock market developments in the short and long term. This discovery may
eventually help investors make better decisions. Around 2014, The Associated Press Twitter account got hacked and a fake tweet was sent
out, saying Barack Obama was injured. As a result of this fake tweet the stock
market went down for 1% for Dow Jones index. How much is that? More
than 100 billion US dollars got swept out. The focal point of this study is
actually to find out whether there is additional information we can extract out of social media in order to better predict stock market performance. Perhaps we can extract information that goes above and beyond what has been captured
within the public news. At the very beginning we collected about 21 weeks of tweet data and extracted all the information in all the tweets that
mentioned a top S&P 100 firm. We had to determine the sentiment being
expressed in those tweets: positive, neutral and negative,
indicating that some investor wants to buy, keep or sell stocks that they hold for a given company. What we demonstrated is, indeed, based on that Twitter information, with the value extracted out of this tweet information, expressed in
sentiment, you can make smarter and better trading strategies and earn
excess returns. Even when we take into account the transaction costs, as well as the fixed costs for running this exercise, we can see that there’s still
excess returns, compared to the market performance. The implication goes way
beyond just the financial industry. A lot of what we do and what we say, both
in online and offline environment, has been digitised. We can learn from this
digitised human behavior how individuals make their
decisions, and also firms can learn from this to make better decisions
that can target and also personalise their services and product at the
individual level in order to achieve better performance.

Leave a Reply

Your email address will not be published. Required fields are marked *

Back To Top